[SeoulTech x Dong-A.com Joint Project] Seoul National University of Science and Technology (hereinafter referred to as SeoulTech) operates various support programs to aid the growth of startups, including preliminary and early startup packages, makerspaces, and global collaboration. Furthermore, in partnership with Dong-A.com, it provides global news to assist startups in overseas promotion and expansion, introducing promising deep-tech startups both domestically and internationally.

DXR develops manufacturing process automation solutions using AI vision, robotics, and digital twin technology. The core solution is the AI inspection automation solution RoV (Robot Vision). RoV attaches a camera and lighting to the end of a robotic arm to scan all surfaces of a product in real-time. This allows it to be easily applied to mixed-flow processes that produce multiple types of products, non-standard products, and various products on a single production line. Additionally, it develops SynthVision, a virtual defective product data generation toolkit for AI model training. SynthVision generates defective product data from images of good products, helping AI-adopting companies reduce the time and cost of building data sets.

After introducing RoV 1.0 in December last year, DXR conducted Proof of Concept (PoC) with three mid-sized companies related to automotive parts and proceeded with supply contracts after its performance and effectiveness were recognized. SynthVision was provided to AI-adopting companies through government-supported projects and received positive feedback. DXR aims to contribute to the resurgence of domestic manufacturing through solution enhancement, business expansion, and entry into the global market.

An interview was conducted with Eunho Kim, CEO of DXR, regarding DXR, RoV, and SynthVision.


Providing Manufacturing Process Automation Solutions, DXR

IT Dong-A: Hello, CEO Eunho Kim. Please introduce yourself.

Eunho Kim: Hello, I am Eunho Kim of DXR. I graduated with a master's degree in Artificial Intelligence from Sungkyunkwan University, with AI vision as my main research area. During my student days, I researched AI vision for mobile robot platforms in a control robot lab. After graduation, I worked as a researcher at LG Electronics Production Technology Institute and Hyundai Mobis, participating in the development of secondary battery vision inspection solutions and lane recognition solutions. I joined DXR in 2023.

IT Dong-A: What led you to join DXR?

Eunho Kim: Throughout university and graduate school, I experienced various process automation projects involving mobile robots and manipulators (robot arms that manipulate and move objects). In the field, I realized the competitiveness of process automation. I often discussed with my university peers the lack of standout startups in the robotics and manipulator fields, which led us to create an innovative company in this area. This motivated me to join DXR. After its establishment as a corporation in 2022, DXR made various attempts, and in 2023, I joined with my lab colleagues to officially start the business.

IT Dong-A: What kind of company is DXR?

Eunho Kim: DXR is a startup supplying manufacturing process automation solutions. The company name stands for Digital Transformation Research, signifying our provision of digital transformation-related solutions. Our mission is to 'create a future of growth with customers through innovative technology,' and our vision is to 'realize a future where humans and robots coexist,' developing automation solutions related to manufacturing processes. We have applied for eight patents related to AI, robotics, and digital twins, with one patent registered and one international patent cooperation treaty (PCT) application, continually securing patents.


Camera and Lighting Attached to Robotic Arm, RoV

IT Dong-A: Please introduce DXR's main solutions and features.

Eunho Kim: The manufacturing industry is shifting from mass production of a few types to small-scale production of many types. However, existing automation solutions struggle to adequately respond to this change due to limitations such as securing quality data and the additional costs and time incurred when changing products. We are developing automation solutions that match manufacturing trends using AI vision, robotics, and digital twin technology.

The core solution is the AI inspection automation solution RoV. RoV is a solution that inspects the exterior quality of products using AI vision and robotics technology. Its main feature is the design that attaches a camera and lighting to the end of a robotic arm. By setting the optimal scan path for the product, the robot moves and scans all surfaces of the product in real-time.

This allows it to be applied to mixed-flow processes that produce multiple products on a single production line. Most existing AI vision inspection solutions are fixed. Since cameras and lights are fixed in specific positions, changing products requires additional costs and time to reconfigure the cameras and lights, making them unsuitable for mixed-flow processes. However, RoV can be easily applied to mixed-flow processes by simply changing the robot arm settings, significantly reducing reconfiguration costs and time. Actual PoC results showed about a 70% reduction in reconfiguration costs compared to existing solutions.

Additionally, RoV can inspect products moving on a conveyor belt in real-time. Fixed solutions often require stopping the belt for inspection, which can disrupt production volume or process cycle time. RoV continuously captures images while the robot moves, not affecting production volume or cycle time. It is also effective for inspecting non-standard products, as the robot captures all surfaces of the product, detecting defects that fixed inspection solutions might miss.

We are also developing AI Picking and DX Universe. AI Picking is a solution that picks up objects from a specific location and moves them to another. It accurately picks and moves the required product even when various objects are mixed. AI Picking has completed PoC and is about to sign a supply contract.

DX Universe is a robot control automation solution that generates robot simulation screens using digital twin technology, allowing easy setting of robot arm movements. This enables non-experts to easily set the position, movement, and desired tasks of the robot arm. DX Universe is being prepared to be offered as a total manufacturing automation solution along with RoV and AI Picking.


Generating Defective Data from Good Product Images, SynthVision

IT Dong-A: I heard that you are also developing SynthVision besides process automation solutions. What is SynthVision?

Eunho Kim: SynthVision is a toolkit that virtually generates defective product data. Our solution is AI-based, requiring data of both good and defective products to train AI models. However, collecting defective product data is not easy.

SynthVision uses generative AI and 3D deformation technology to apply scratches and dents to images of good products, creating virtual defective product data. Size and position can be freely adjusted. Unlike other solutions, SynthVision's distinction is that it generates defective product data using only images of good products, eliminating the need for separate defective product data. It also actively incorporates the opinions of field workers to enhance accuracy and efficiency.

SynthVision was launched in the first quarter of this year, and PoC confirmed that it reduced the cost and time of building AI model training data sets by over 50% and the time to adopt AI vision inspection solutions by over 90%. We are currently enhancing it by expanding defect types and improving data accuracy.

From PoC Completion to Supply

IT Dong-A: Please tell us about the progress of your business so far.

Eunho Kim: For RoV, after the release of version 1.0 in December 2024, we conducted PoC with three mid-sized companies related to automotive parts. As a result, we confirmed over 30% labor cost savings and over 40% improvement in defect detection rates. Recognized for this, we proceeded with supply contracts and are negotiating additional contracts.

Automotive parts processes often involve multiple products and mixed-flow processes, making AI vision inspection challenging to adopt. However, RoV's high inspection accuracy, process automation, labor cost savings, and reduced worker fatigue have resulted in high customer satisfaction. Some companies have been considering inspection automation for 10 years, and this is the first time it has led to actual supply.

For SynthVision, we completed actual sales through government-supported projects. Companies adopting AI inspection solutions have responded positively, noting the ease of training AI models.

Additionally, we secured seed investment from KingoSpring in 2024. We are participating in various government support projects, including the TIPS (Tech Incubator Program for Startups) led by private investment, Didimdol R&D, and early startup packages. Our current location at Seoul AI Hub provides ample space to accommodate robots, enhancing work efficiency.


IT Dong-A: You are currently receiving support from SeoulTech's early startup package. What support have you received?

Eunho Kim: Through SeoulTech's early startup package program, we received commercialization funds and various growth support programs. We had opportunities to participate in various programs such as 1:1 meetups with investors and open innovation programs, receiving practical help necessary for company growth.

IT Dong-A: Lastly, please tell us about DXR's future plans.

Eunho Kim: We plan to introduce RoV version 2.0 within this year. We are developing it by incorporating customer feedback on shooting speed and volume, lighting optimization, robot control algorithm optimization, and AI model performance enhancement, expecting overall performance improvement. Additionally, we plan to expand the scope of manufacturing process automation to areas such as displays and food and beverages through the release of manufacturing automation solutions like AI Picking and DX Universe.

We are also preparing to secure pre-Series A investment. Through this, we plan to secure additional R&D personnel and enhance solutions, focusing more on securing technological competitiveness. Alongside this, we plan to establish overseas branches next year for full-scale entry into the overseas market, currently targeting the large North American market. Ultimately, we aim to contribute to the resurgence of the domestic manufacturing industry by building a total smart factory solution that aligns with current manufacturing trends.

IT Dong-A Reporter Han Man-hyuk (mh@itdonga.com)

Popular News